An Analysis of Social Networks Analysis in Online and Face-to-Face Bridge Communities

1. An Analysis of Social Networks Analysis in Online and Face-to-Face Bridge Communities Alexandru Iosup Vlad Posea, Mihaela Balint, Alexandru Dimitriu Politehnica University of Bucharest, Romania Parallel and Distributed Systems GroupDelft University of Technology Presented by Dick Epema. (Many thanks from the BridgeHelper team.)

2. What’s in a name? Massively Social Gaming (online) games with massive numbers of players (100K+), for which social interaction helps the gaming experience • Virtual worldExplore, do, learn, socialize, compete+ • ContentGraphics, maps, puzzles, quests, culture+ • Game analyticsPlayer stats and relationships

3. Sources: MMOGChart, own research. Sources: ESA, MPAA, RIAA. MSGs are a Popular, Growing Market • 25,000,000 subscribed players (from 150,000,000+ active) • Over 10,000 MSGs in operation • Market size 7,500,000,000$/year

4. Social Networks: Buzzword? Science? • Social Network=undirected graph, relationship=edge • Community=sub-graph, density of edges between its nodes higher than density of edges outside sub-graph

5. Sources: CNN, Zynga, 2010. Source: InsideSocialGames.com FarmVille, a Massively Social Game Key advantage over market:Use [Social Network] analysis to improve gameplay experience Zynga CTO

6. Agenda • Background on Massively Social Gaming • Bridge, the Running Example • Research Question • Addressing the Research Question • Conclusion

7. Bridge, A Traditional Team Card Game • Bridge as traditional card game • Hand=one “game” • 2 pairs (4 players) play hands (bidding + play) • Duplicate bridge • Team=2 pairs at separate tables • Same hand at every table • Same team plays opposite ends • Eliminates luck • Only team game at last World Mind Sport Games, Beijing, 2008

8. Bridge, a Special Use Case of SocNets? • Similarities • Online and Face to Face • Complex agreements between partners (like a social partnership) • A good pair forms in a very long period of time (like a social …) • Differences • Adversarial context, not only cooperation and ‘friendship’ • Gaming social networks have no strict definition of relationship (‘played once’ vs ‘day-to-day partner’) • Links in the network not specified precisely

9. Research Question: What are the Characteristics of Bridge Communities? • Study the activity and socnet characteristics of online and face-to-face bridge communities • Why is this interesting? • Unique type of social network? (new knowledge) • Unique type of social gaming network? (new knowledge) • Use results to develop new services (matchmaking, rating) • Use results to improve online game operations (player retention) • “Real-world” applications: other social network results applied in economics; adversarial settings good for management and psychology studies; etc.

10. Agenda • Background on Massively Social Gaming • Bridge, the Running Example • Research Question • Addressing the Research Question • Method • Data • Analysis Results • Conclusion

11. Analysis of BBOFansMethod • Gather data from online and face-to-face communities • Data: who played with or against whom, and when? • Analyze player activity levels [see article] • Transform the play data into G=(V,E), V=set of players, E=set of social relations. • Investigate social relations based on play relationships • Analyze properties of graph G • Traditional socnet analysis, e.g., community detection • Player type analysis • Use face-to-face data to guide analysis of online data

12. 1. Gathered DataBBO (Fans): Massively Social Gaming • Bridge Base Online (BBO) http://www.bridgebase.com • Largest online bridge platform, free to play • 1M active players, also attracts many professional players • Friends and enemies, filtering by skill and nationality • No advanced social networking features, e.g., No Friends-of-Friends • BBO Fans http://www.bbofans.com/ • Uses BBO for actual gameplay • BBO Fans community included in BBO • Better social network facilities • Community tools: awards, ranking, rated tournaments, etc. Vlad Posea, Mihaela Balint, Alexandru Dimitriu, and Alexandru Iosup, An Analysis of the BBO Fans online social gaming community, RoEduNet International Conference (RoEduNet), 2010 9th.

13. 1. Gathered DataLocomotiva: Face-to-Face Bridge • Locomotiva http://www.locomotiva.ro • Typical of many large clubs around the world [see article] • Large bridge community, free to play • ~275 active players, also attracts many top players • 4 tournaments per week, 15 bigger tournaments per year • 20-60 people per tournament, ~4h/tournament • Games/Tournaments recorded as participants and results Vlad Posea, Mihaela Balint, Alexandru Dimitriu, and Alexandru Iosup, An Analysis of the BBO Fans online social gaming community, RoEduNet International Conference (RoEduNet), 2010 9th.

14. 1. Gathered DataDatasets • Face-to-face bridge data • Created real-world club management software • Locomotiva data • Online bridge data • Created domain-specific web crawler • BBO + BBO Fans data (BBO Fans included in BBO)

15. 3. Transform Data into Social LinksWhat is a Link? A New Framework • Main idea: Two players have a social relationship if they relate strongly through play • They are at the same place at the same time • They have played together or against each other • A number of hands • A number of sessions (all hands in one sitting) • They are part of the same team • Can extract social relationships from our datasets • Single criteria + thresholds • Multi-creteria + multiple thresholds

16. 3. Transform Data into Social LinksResults of Transformation Maximum modularity • Method • Different criteria + thresholds • Validate for Locomotiva using human experts (from the club) • Present extracted communities to expert • +1 if regular partners in same community, etc. • Validated validators via maximum modularity (Q) • (P+>=200) OR (S+>=8) • Played hands as partners (P+) • Sessions as partners (S+) Mean community size # of communities Non-isolated nodes

17. 3. Transform Data into Social Links/4. Analysis of GNormalization + Analysis results • Normalization • Threshold values valid for a given community size • Played hands and sessions are cumulative in # of weeks • For Locomotiva: 50 weeks • For BBO: 5 weeks • For BBO • P+ >= 20 (200 x 5 / 50) • Obtained modularity Q = 0.43 (same as for Locomotiva) • 4,375 communities, 90% of which have at most 4 players

18. 4. Analysis of GPlayer Types • Community Builderplays many hands withmany other players • Community Memberplays mostly with a few community members • Faithful Player1-2 stable partners • Random Playerno stable partner Goal for the future: Reduce # of random players in Face-to-Face bridge

19. Agenda • Background on Massively Social Gaming • Bridge, the Running Example • Research Question • Addressing the Research Question • Conclusion

20. Massively Social Gaming • Million-users, multi-bn. market • Content, World Sim, Analytics Current Technology Our Vision • Complete game mechanics • Basic social network tools • Makes players unhappy • Many starters quit • Social Network Analysis +Applications = BridgeHelper Ongoing Work • More analysis • Ranking • Matchmaking The Future • Scalability, efficiency • Happy players

21. Thank you for your attention! Questions? Suggestions? Observations? More Info: Alexandru IosupA.Iosup@tudelft.nlhttp://www.pds.ewi.tudelft.nl/~iosup/ (or google “iosup”)Parallel and Distributed Systems GroupDelft University of Technology • http://www.st.ewi.tudelft.nl/~iosup/research_gaming.html • http://BridgeHelper.org (soon)